Case Study: Train the model on some custom dataset with manual resolution adjustments, and then test the trained model on a new but similar image

Ref High Resolution Dataset	SRCNN Predictions from Low Resolution (0.5 x)
Used in Training	Prediction
Not used in Training	Prediction

Case Study: Using a pretrained model Nina_SR model (trained on Div2K dataset on 300 epochs), and testing on new images

Ref High Resolution Dataset	Nina_SR Predictions from Low Resolution (0.5 x)
Not used in Training	Prediction
Not used in Training	Prediction

Case Study: Improving Resolution Improves 3D rendering result (COLMAP Dense Reconstruction Example)

Low Resolution Dataset

lowres_video.mp4

High Resolution Dataset (x2)

highres_pred_video.mp4

Reference

https://github.com/isaaccorley/pytorch-enhance

https://github.com/Coloquinte/torchSR